Class: RubyCoded::Tools::ExecutionPipeline

Inherits:
Object
  • Object
show all
Defined in:
lib/ruby_coded/tools/execution_pipeline.rb

Overview

Standardizes the internal execution flow shared by side-effecting tools:

1. validate input (project-relative path if given)
2. resolve to an absolute path inside the project root
3. optionally forbid operating on the project root itself
4. execute the tool's business logic (via block)
5. normalize filesystem errors into a structured response

Policy, risk assessment and user approval intentionally live outside this pipeline: they are handled by Tools::ExecutionPolicy and by the bridges (BridgeCommon::ToolFlow), which sit above tool execution.

Instance Method Summary collapse

Constructor Details

#initialize(project_root:) ⇒ ExecutionPipeline

Returns a new instance of ExecutionPipeline.



16
17
18
 
# File 'lib/ruby_coded/tools/execution_pipeline.rb', line 16

def initialize(project_root:)
  @project_root = File.realpath(project_root)
end

Instance Method Details

#call(path: nil, forbid_root: false) ⇒ Object

Runs the pipeline. Yields the resolved full_path (or nil if path was not provided) to the caller-supplied block, which returns the tool result — either a success value (string/hash without :error) or an error hash of the form { error: "..." }.

Any SystemCallError raised inside the block is normalized to an error hash.



27
28
29
30
31
32
33
34
35
36
37
38
 
# File 'lib/ruby_coded/tools/execution_pipeline.rb', line 27

def call(path: nil, forbid_root: false)
  full_path = nil
  if path
    full_path = resolve_and_validate(path)
    return full_path if full_path.is_a?(Hash)
    return { error: "Cannot operate on the project root" } if forbid_root && full_path == @project_root
  end

  yield(full_path)
rescue SystemCallError => e
  { error: "Filesystem error: #{e.message}" }
end